sankethkatta/simpy

Event Scheduling Simulation Engine

simpy - Event Scheduling Simulation Engine

simpy is an event scheduling simulation engine implemented in python.

Introduction

simpy consists of 2 main components Events and Edges. Simulations can be thought of as node and pointer diagrams:

+-------+               +-------+
| EVENT |---- EDGE ---->| EVENT |
+-------+               +-------+

Events change state variables, and return Edge(s). An Event can return multiple Edges.

Edges check conditions and return an Event. They can also schedule the Event at a delay. An Edge must always return exactly 1 Event.

The process launched by a "Run" Event. This first Event sets off the chain of Event Edge calls.

Variables

Simpy operates mainly on a global namespace. By default a function does not take in any parameters. Therefore, the preferred method is to create an object that can be globally accessed across all functions, Simpy's Variables class.

# Instantiate the class
var = Variables("QUEUE", "SERVER")

# Access or Set the variables
var.QUEUE = 0

Event

Events in Simpy are simply functions with a decorator, which let the engine know that the function is to be treated as an Event. The decorator takes 1 parameter name this is the name of the Event that will be displayed on outputs traces.

@Event(name="Run")
def run():
   var.QUEUE = 0
   var.SERVER = 2
   return run_to_enter

Edge

Edges are also simply functions with a decorator. The Edge decorator has 2 optional parameters,delay and priority. The default delay is 0 and default priority is 5 (lower number is higher priority).

@Edge(delay=0)
def leave_to_start():
    if var.QUEUE>0:
        return start

Run

Finally to run the simulation, the engine needs to know the total runtime, the first function to be called to set off the simulation (usually called run) and the name of the Variables instance being used.

RUN_SIMULATION(RUNTIME=50, RUN=run, VARIABLES=var)